Numpy array assignment

Question

I've written a very simple python numpy code. It have a strange behavior...

from numpy import *
# generate 2 array with 15 random int between 1 and 50
pile = random.randint(1, 50, 15)
pile2 = copy(pile)

print("*** pile2",type(pile2),pile2)
print("tab with fixed values ")
tmp2=array([155,156,157,158,159])
print("tmp2",type(tmp2),tmp2)
pile2[:5]=tmp2
print("pile2",type(pile2),pile2)

print("*** pile",type(pile),pile)
print("flip a part of pile and put in an array")
tmp=pile[4::-1]
print("tmp",type(tmp),tmp)
pile[:5]=tmp
print("pile",type(pile),pile)

When I run this script, it return :

*** pile2 <class 'numpy.ndarray'> [20 23 29 31  8 29  2 44 46 17 11 47 29 43 10]
tab with fixed values 
tmp2 <class 'numpy.ndarray'> [155 156 157 158 159]
pile2 <class 'numpy.ndarray'> [155 156 157 158 159  29   2  44  46  17  11  47  29  43  10]

Ok! pile2 become something like "tmp2[] and pile2[6::]", but for the second...

*** pile <class 'numpy.ndarray'> [20 23 29 31  8 29  2 44 46 17 11 47 29 43 10]
flip a part of pile and put in an array
tmp <class 'numpy.ndarray'> [ 8 31 29 23 20]
pile <class 'numpy.ndarray'> [ 8 31 29 31  8 29  2 44 46 17 11 47 29 43 10]

tmp [ 8 31 29 23 20]

pile [ 8 31 29 31 8 29 2 44 46 17 11 47 29 43 10]

Oh! There is problem with assignement ! What's happens ?

I am not able to duplicate this behavior. It is possible that there is an error elsewhere in your code. By the way, try to avoid using from [module] import * as it can cause namespace-related problems. — mechanical_meat
– mechanical_meat, Commented Mar 15, 2012 at 16:12

segasai · Accepted Answer · 2012-03-16 01:14:42Z

2

I can confirm the behaviour with numpy 1.3.0. I guess this is indeed an old bug. And this:

pile[:5]=tmp.copy()

solves the issue.

answered Mar 16, 2012 at 1:14

segasai

8,5581 gold badge31 silver badges40 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

HYRY · Accepted Answer · 2012-03-16 00:56:42Z

0

Because tmp is a view of pile, when you use it to set the content of pile it can cause problem. I am using NumPy 1.6.1, and cannot duplicate this, so maybe this is fixed in the newest version. If you are using old version, you can try:

tmp=pile[4::-1]
pile[:5]=tmp.copy()

answered Mar 16, 2012 at 0:56

HYRY

97.8k28 gold badges197 silver badges192 bronze badges

Collectives™ on Stack Overflow

Numpy array assignment

2 Answers 2

Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related